1. Technical Field
The present disclosure relates to apparatuses and methods for processing an image to remove a background from an input image.
2. Description of the Related Art
Image recognition technology to acquire images and recognize the acquired images has been applied to a variety of fields. For example, the image recognition technology has been applied to touchscreens to judge a user's touch based on sequentially acquired images of the user's hand, to automobiles to recognize a gesture of a user and provide predetermined feedback in accordance with the recognized gesture of the user, and to the like.
For image recognition, there is a need to remove a background from an object to be recognized. The background may be removed using various methods such as a static removal method using a pre-stored background image, a difference image method by acquiring and using a plurality of images, and a depth map method using a stereo camera.
According to the static removal method, a background is removed from an object by removing a pre-stored background image from an acquired image. The static removal method has a low degree of adaptability to external disturbing light and a low ability to recognize the object in low-light environments.
According to the method using a depth map, a background is removed based on a depth map calculated using stereo images simultaneously acquired by a stereo camera. However, since a lot of computations are required to calculate the depth map, high performance hardware is required to calculate the depth map. Also, the method using the depth map has a low degree of adaptability to external disturbing light.
According to the difference image method, a background is removed from an object by using a difference between an image acquired when a lighting unit is turned on and an image acquired when the lighting unit is turned off. The difference image method has a high degree of adaptability to external disturbing light and shadow removing effects. However, since accuracy of removing the background according to the difference image method is affected by a shutter speed of a camera, a high quality camera is required to increase accuracy.